Forward-deployed engineers expose the services core of AI (McKinsey)

The renewed interest in forward-deployed engineers comes with an awkward admission from McKinsey: the title may be a buzzword for work consultants have offered for years.

A forward-deployed engineer sits inside the client, in its systems and data, building software alongside the people who will use it. McKinsey's explainer concedes the line between an FDE and a traditional consultant is still blurry, and one of its own leaders says the roles are merging. Their descriptions of the work are telling: leave a product behind, own an outcome rather than a recommendation, and change the client's business process along with its software.

The rebrand matters less than what it reveals. The model labs and software vendors have discovered that AI does not deliver value off the shelf. It has to be fitted to a specific process, specific data and a specific set of exceptions, and that fitting is skilled labour on site. As one of McKinsey's leaders puts it, you cannot just go in, deliver insights and walk away.

The strongest case for software economics is that this labour is temporary: embedded teams resolve the early integration problems, then standard products take over. That may happen, but buyers should not pay today as though it already has.

If every serious deployment needs embedded engineers, AI looks less like software with software margins and more like a services business with a software core.

For buyers, the practical move is to treat FDE capacity as a scarce input in vendor selection. The best model with no one to fit it will lose to a decent model with a team in the building.

Source: McKinsey Explainers, "What is a forward-deployed engineer?" https://lnkd.in/ghgvKGSp

#AIAdoption #Consulting #Strategy